3 Exercise: Project Progress - Video Tutorials & Practice Problems

<v ->All right, back to you.</v> Time to return to your course project. So remember what your project is. So you're gonna be developing an AI strategy for your chosen organization or solving some kind of specific problem using the process that we laid out for you. So what are we gonna do this time? Well, now we're gonna go deeper on what your first idea is for AI. So what is your current process? So can you document it? Is it documented already? And does it depend on human judgment or is it something that has an objective outcome? That's what we really want you to think about. Now, sometimes you get both. Sometimes you have both an objective outcome and human judgment, so that's okay if it's both, but you really need to think about how you get that outcome data. How will you know what the right answers are for each situation that you're trying to solve the problem for? And how much data do you have? Do you have a lot of historical data already and you can just put it to work or do you need to start to think about how you're gonna collect that data? How are you gonna change the process so that you start now collecting the data you need in order to solve the problem with AI? And what is the value of the improvement? We talked about how there's all these different forms of AI. Some of them are more expensive than others. You have to really get serious about knowing what the value is. We tried calculating the ROI as part of your project previously, but what I want you to be thinking about now is, was that really correct? Can you really make the case that that's the value of the improvement? So go deeper on this first idea and try to flesh it out a little bit more and figure out whether you've really done the previous exercises the way you need to. What is the current process? What kind of data are you using and how much do you have? And are you really able to use all of that to show the value of what the improvement would be? And then start thinking about what kind of AI do you need? Can you solve it with an embedded approach? It's just part of a piece of software. You either have to get that piece of software or you have it already and you have to start using it in this new way. Or is it a data-driven model? Is it something where you can get it in a piece of software, but you're gonna have to do some type of IT project, some type of integration project that allows this software to collect your data in order to train its model and do its magic, or do you have a much bigger job in front of you? Do you need type of specialized approach where you have to bring in a data scientist and you have to create your own model, not just your own data, but you need the data scientists to do the feature analysis. They have to figure out what characteristics are important. And they have to create the model. That would be a much bigger proposition. And hopefully, you don't have this huge problem where you have to create a whole custom approach where you're actually trying to figure out new tools and AI to use. That would be relatively rare, but do give some thought as to what kinds of AI are involved. And it could be there's more than one. It could be part of your problem can be solved by something simple, but you've got a bigger problem or you have a problem that you'd like to solve more deeply by going to one of the higher approaches. And so that's okay as well. Think a little bit about budgeting at your company. How does it work where you are? How do people get budget? How do they get projects approved? And is this process gonna work for your AI project? If you remember, there's all sorts of things that you're doing along the way where you're basically experimenting. You're kind of introducing AI to the process and it might not be showing a lot of value at first. Is that something that really works with budgets at your company or do they need you to really have some tightened up process where people are saying, hey, well, I'm gonna have this ready by May and give me the money and it'll be done by May. You might not be able to make that kind of promise because it may take you several months of experimentation before you start to figure out how to solve the problem well enough that you can actually justify spending all this money on it. It could be that in the beginning, you're not even gonna solve the problem as well as the people do. So you have this lovely situation where not only is it not as good as what people were doing, but it's more expensive. It's like, yeah, that's not where we wanna end up. And so it could take you a while of experimentation before you start to solve the problem faster, cheaper, better, and then you can start to see the value out of it. So your budgeting process might be one where you're looking for more of a process where couldn't they just make some kind of bet on an experiment and say, hey, we'll give you this amount of money every month for three months or four months or five months and then we'll take a checkpoint and see where you are, rather than saying, I need to get all of the money now and I have to know exactly how long it's gonna take. And so think about whether your budgeting process needs to change in your organization in order for you to tackle this kind of project. And as always, keep your decision maker in mind. Who's approving the plans? Make sure everything you're doing as you work on this project is targeted towards persuading that person. If you know that person really well, what drives them? What are the kinds of things they would think about? You need to make sure that you're appealing to those things so that your project gets approved. So let's summarize all the things you're gonna do right now. You're gonna continue with your plan by analyzing your first AI initiative for feasibility. Think deeper about what the process is, what data you need, whether you have that data, and then what you wanna do is determine what approach you're gonna use. So can it be embedded, data-driven, specialized? What is it that you need to do? Each of those types of projects are a different level of investment, a different timeframe, and also a different level of risk, and consider whether your budgeting process really accommodates your ability to be able to experiment and mitigate that risk, or does your budgeting process maybe need to change a little bit? If you have a very tight budgeting process where people think that everybody can promise exactly what the outcome will be when, that might not be really a fortuitous situation to be in to try to do AI projects, especially when you start to get up into those specialized approaches. But even for data-driven, that can be hard. It might work for embedded. And so think about your budgeting process and whether you need to really get some kind of dispensation for the normal process in order to be able to do what you wanna do. So go after it. This is your chance to really make this your own. So take a look at your project and try and do these things to really focus on improving your plan.